
Most AI tools wait. You open a chat, type a request, and get an answer. Claude Duet flips that around. It’s an AI agent that watches your work, tells you when something needs attention, and keeps going on tasks even after you’ve closed the app.
Anthropic opened a research preview of Claude Duet on October 21, 2026. Here’s what it is, how it works, what’s changed since launch, and what to know before you join the waitlist.
What is Claude Duet?
Claude Duet is a standalone app where Claude acts as a proactive agent. It’s built for professionals and small business owners who have more to track than any one person can keep up with: inboxes, calendars, invoices, customer messages, deadlines.
It’s powered by Claude Opus 5.5, so it’s a new app built on an existing model, not a new model. Anthropic says it designed Duet around judgment: deciding what needs you now, what can wait, and when to check with you first.
You can use it as a standalone app on iOS and Android, or on the web at claude.ai.
What does it actually do?
A morning brief. Each day, Duet sends a short summary of the most important items across your connected apps.
Background monitoring. During the day it keeps an eye on those apps and only messages you when something deserves your attention. It checks every few hours rather than continuously, because constant checking would use up your plan’s limits much faster. You can tell it what you want to hear about and how often.
Tasks of all sizes. You can hand it something small, like scheduling a meeting, or something bigger, like a financial or sales forecast, a dashboard, a prototype, or a web app.
One ongoing thread. You talk to it by typing or by voice, in a single conversation. A quick reply and a week-long project start in the same place.
It carries over your Claude setup. Your memory, skills, and connected apps from your Claude account come with you, so there’s little to configure.
It keeps working when you step away
Duet runs on its own virtual machine with a browser, so your phone or laptop doesn’t need to stay on. Describe a task in a short message, and it works out which apps and steps are needed. You get a notification when it’s finished or when it needs your review.
For the kind of person Duet is aimed at, that matters. You can hand off your to-do list before leaving for the day and find the results waiting for you later.
You stay in charge of the big decisions
An agent that can send emails and spend money needs guardrails. Duet reads, drafts, and prepares work on its own, but before it sends a message, makes a payment, signs a document, or deletes anything, it shows you what it’s about to do and waits for your approval.
Some things are off-limits even with your approval, such as trading securities or moving money between your accounts. You can also disconnect any app it has access to at any time.
Safety, privacy, and security
Anthropic highlights several layers of protection, and published a separate article on them two days after launch:
Credentials stay hidden. Duet never sees your passwords, browser cookies, or sign-in tokens. When a site asks for a password, you type it into a form inside the Claude app, and credentials only pass through separate machines the model can’t access.
A separate check on actions. Approvals and safety checks run outside the model that does the work, so instructions hidden in a web page or email can’t approve their own actions. Actions are also screened by the same auto mode classifier that guards Claude and Claude Code against steps you didn’t ask for.
Sensitive-information classifier. A dedicated classifier checks each message before it goes out. Requests to share the strictest tier of data, such as credit card numbers and government ID, are rejected. Other sensitive details, like health or personal finance information, go out only when you explicitly tell Duet to send them. Anthropic acknowledges that these checks can miss things or flag harmless messages during the preview.
Prompt-injection defenses. Protections against malicious instructions in pages and downloads draw on Anthropic’s browser-use research.
Testing. Duet was tested by red teams and an external penetration-testing firm, and it’s covered by Anthropic’s public bug bounty program.
Anthropic’s support materials still advise reviewing drafts and approval requests, since prompt injection remains a risk for any agent that reads the web.
Connected to the tools you already use
Duet works with the 6,000+ connectors and plugins in the Claude Marketplace. Examples from Anthropic’s announcement:
Tool | What Duet can help with |
|---|---|
Gmail and Slack | Find messages you missed and help you reply |
QuickBooks and Stripe | Spot unpaid invoices and draft reminders |
Gusto | Prepare and run payroll |
Draft a job posting and publish it after you approve | |
Docusign | Prepare agreements and send them for review and signature |
HubSpot | Draft follow-ups using contact and deal history |
Canva | Turn a product launch into a social campaign |
Shopify | Set up, design, and manage a store |
For purchases, Duet can use Link (Stripe’s digital wallet), which approves the spend with you and pays with a virtual card so your real card number isn’t shared. It can also check out through Shop Pay. Signing in to sites with logins saved in 1Password was listed as coming soon at launch.
How early testers use it
Anthropic shared a few stories from early access:
Jess Provencher, broker and owner of Pro Homes, got a morning rundown and one day found that Duet had flagged a property showing booked for the wrong day, without being asked.
Mya Dalton, a brand strategist, had a client call prepared for her. Duet noticed the call fell right before her flight and offered to move it.
Amin Ebrahimi, who owns Shahs of Kabob in South Florida with about 110 employees, used it for accounting, customer support, and marketing, and worked through invoices store by store.
Inside Anthropic, team members use it to talk through to-do lists on walks and to schedule interviews, build prototypes, and ship code.
What’s new since launch
Duet is moving quickly. These updates were shared by Anthropic team members on X and relayed by outlets that track them, so check Anthropic’s own pages for the latest:
Claude Code sessions (early November). Duet can now check on Claude Code sessions, whether they’re running on your computer or in the cloud, so you can see what’s running, approve changes, and send updates from your phone.
Waitlist invites (November 12–13). Karan Ahuja, who leads the Duet team, said that every US Pro, Max, and Ultra subscriber who was on the waitlist as of November 12 has been invited. Joining the waitlist is still how you get in if you haven’t signed up. He also said the team aims to open Duet to more countries outside the US soon.
Usage resets. Anthropic has given new users usage-limit resets to try Duet, and the terms have changed since launch, so check the Duet help article for the current ones.
The fine print
This is an early preview, and Anthropic is upfront about its limits:
It can misjudge. Duet may misread a thread or get the urgency wrong, and it can miss something that arrives between check-ins.
Work comes first. Personal errands, such as booking tickets, aren’t as strong yet.
Limited visibility at launch. Duet couldn’t yet show you its browser, though you can see a task list of what it’s working on and what needs you.
It’s a separate app. For now, Duet is not part of the main Claude app. Anthropic says what it learns will shape how the two come together.
It shares your usage limits. Background checks count toward your plan, and larger jobs like financial models or decks take more model turns.
Who can get it?
Duet is a research preview available by waitlist to Pro, Max, and Ultra subscribers in the US. Team and Enterprise plans aren’t included yet, and it isn’t available outside the US.
You can join the waitlist at claude.com/product/claude-duet.
The takeaway
Claude Duet is Anthropic’s entry in the growing field of proactive personal agents, alongside offerings like Meta’s Muse and OpenAI’s Dots. The bet is that people don’t need another chatbot to prompt. They need something that notices what matters, handles the routine, and stops to ask before anything consequential happens.
Whether it earns that trust will depend on how well its judgment holds up in real inboxes and real businesses. The research preview is where Anthropic plans to find out.
Sources
Press coverage from Engadget, Help Net Security, and The New Stack
Claude Duet is a research preview, so features, availability, and plan eligibility may change. Facts in this post were checked on November 17, 2026.

Comments(0)
Join the discussion securely with your ZyVOP account.
Continue on ZyVOP to CommentNo comments yet. Be the first to comment!